Enzymatic basis of macrophage activation. Kinetic analysis of superoxide production in lysates of resident and activated mouse peritoneal macrophages and granulocytes.

To compare the kinetics of the O-2-generating enzyme in nonactivated and activated macrophages and granulocytes from the mouse peritoneal cavity, we sought conditions in which the activity of this enzyme in cell lysates was comparable to that in intact cells. Pretreatment of macrophages with 10 mM diethyldithiocarbamate inhibited endogenous superoxide dismutase by 70% and enhanced O-2 secretion up to 15-fold, so that it was comparable to H2O2 secretion. O-2 secretion was terminated by detergent lysis and reconstituted by addition of NAD(P)H to the lysates. Optimal detection of O-2 production in lysates depended on prior stimulation of the respiratory burst, lysis with 0.05% deoxycholate rather than any of 4 other detergents or sonication, acetylation of the cytochrome c used as an indicator, and addition of NADPH rather than NADH. Kinetic analysis using NADPH-reconstituted deoxycholate lysates, together with spectra of oxidized and reduced cells, failed to reveal either marked differences in the Vmax of the O-2-generating enzyme or correlations between O-2 secretion and cytochrome b559 content among 5 macrophage populations whose H2O2 secretion ranged from 0 to 365 nmol/90 min/mg of protein. In contrast, the Km of the oxidase for NADPH varied markedly and inversely with the capacity of the intact cells to secrete O-2 or H2O2: J774G8 histiocytoma cells, 1.43 mM; resident macrophages, 0.41 mM; proteose peptone-elicited macrophages, 0.20 mM; casein-activated macrophages, 0.05 mM; NaIO4-activated macrophages, 0.05 mM; and granulocytes, 0.04 mM. These results suggest that macrophage activation, a process that enhances oxygen-dependent antitumor and antimicrobial functions, may equip the cell to secrete increased amounts of reactive oxygen intermediates largely by increasing the affinity of the oxidase for NADPH.

The abbreviations used are: ROI, reactive oxygen intermediates; DDC, diethyldithiocarbamate; MEM, Eagle's minimum essential medium, a variant; HS, horse serum heated at 56 "C for 30 min; EDTA, ethylenediaminetetraacetate; 5% HS-MEM, Eagle's minimum essential medium, a variant, with 100 units/ml of penicillin, 100 pg/ml of streptomycin, and 5% HS; KRPG, Krebs-Ringer phosphate buffer containing 5.5 mM glucose; PMA, phorbol myristate acetate; PBSG, CaZ+,Mg2+-free phosphate-buffered saline containing 5.5 mM glucose; S.D., standard deviation. oxide and hydrogen peroxide (1-4). The superoxide-generating enzyme of the granulocyte is believed to be a plasma membrane-associated flavoprotein (4, 5) that oxidizes NADPH (1) in concert with a cytochrome b669 (6) and an ubiquinone (7). In chronic granulomatous disease (8), decreased ability of granulocytes to produce ROI and thereby to resist infection can result from impairment of NADPH generation (9), absence (10) or abnormality (5, 11) of the cytochrome bmg, or altered affinity of the oxidase for NADPH (12,13). Much less is known about the enzymatic basis for production of ROI in macrophages (14-17). The question holds special interest because macrophages are subject to immunologic activation, which can elevate their capacity to secrete ROI from a level like that of chronic granulomatous disease granulocytes to a level comparable to normal granulocytes (3). The position of macrophages along this spectrum is regulated in part by interferon-y (18). ' The goal of this study was to define, at the enzymatic level, the basis for the marked difference in the ROI secretory capacity of mouse resident peritoneal macrophages compared to activated macrophages and granulocytes. For this purpose we sought assay conditions with lysed cells in which ROI production was as great as with intact cells. The fulfillment of this criterion required changes in reported procedures for estimating macrophage superoxide production. In contrast to the situation with granulocytes, the activity of superoxide dismutase and cytochrome c reductase in mouse peritoneal macrophages, together with the lability of their oxidase after cell disruption, presented major obstacles. We found it necessary to inhibit intracellular superoxide dismutase with DDC (19), to disrupt cells with deoxycholate rather than other detergents or sonication, and to monitor superoxide by the reduction of acetylated rather than native ferricytochrome c (20, 21). Kinetic analysis using these techniques suggested that the increased superoxide-generating capacity of activated compared to resident macrophages can be attributed primarily to an increase in the affinity of their oxidase for NADPH, so that it resembles the affinity of the oxidase in granulocytes. Neither the specific activity of the oxidase nor the specific content of cytochrome b559 appeared to play as great a role.

4305
was from Eastman, acetic anhydride from Fisher, and proteose peptone from Difco. Diff-Quik staining solutions were from Harlem (Gibbstown, NJ). MEM (a variant) was from Flow Laboratories (Rockville, MD). Penicillin, streptomycin, and trypan blue were from Gibco Laboratories (Grand Island, NY). HS was from Sterile Systems, Inc. (Logan, UT).
Cell Preparatiorw-Nelson-Collins strain mice (The Rockefeller University, New York, NY) or ICR mice (Camm Research, Wayne, NJ) of either sex were used at >8 weeks of age. Resident macrophages were washed from the peritoneal cavity with MEM containing 5% HS, 100 units/ml of penicillin, and 100 pg/ml of streptomycin (5% HS-MEM). Where indicated, mice were injected intraperitoneally 3 days before harvest with 1 ml of 5 mM NaIO. or 1% (w/v) proteose peptone in 0.9% NaCl or were injected 5 days before harvest with 1 ml of 6% (w/v) sodium caseinate. Granulocyte-rich populations were collected 6 h after injection of 1 ml of 12% sodium caseinate (22). After centrifugation at 200 X g for 5 min, the cell pellet was treated with cold 0.2% NaCl for 30 s to lyse erythrocytes. After tonicity was restored to 300 mOsm, the cells were passed through nylon mesh (200/inch2) (Tetko, Inc., Elmsford, NY), centrifuged, and resuspended in 5% HS-MEM. J774G8 murine histiocytoma cells were the gift of J. Unkeless, Rockefeller University. In the case of macrophagerich populations used as adherent cells, 0.6 X 106-2 X lo6 cells/70 pl were incubated on 13-mm diameter glass coverslips (Clay Adams, Inc., New York, NY) cleaned as described (18). After 2 h at 37 "C in 5% cOz-95% air, nonadherent cells were removed by agitation of the coverslips in warm medium. The coverslips were incubated overnight in 5% HS-MEM and transferred to 16-mm wells in 24-well cluster trays (Costar Data Packaging, Cambridge, MA). For macrophages in suspension, 4 X 107-5 X lo7 peritoneal cells/7 ml of 5% HS-MEM were incubated in 100-mm diameter plastic dishes (Nunc, Roskilde, Denmark) for 1-2 h at 37 "C (tissue culture dishes from some other manufacturers were unsatisfactory for recovering the adherent cells). Nonadherent cells were removed during 3 washes with warm PBSG (137 mM NaC1, 2.6 mM KC1, 8.1 mM NazHP04, 1.5 mM KHzPOI, 5.5 treated with 10 mM freshly dissolved DDC in PBSG for 60 min at mM glucose, pH 7.4). Where indicated, the monolayers were then 37 'C, with the addition of 10 mM EDTA for the final 15 min. The cells were dislodged by pipetting, washed 3 times in PBSG, and suspended in KRPG, pH 7. 35-7.40. Granulocyte-rich populations were plated directly in PBSG with DDC, recovered 60 min later with EDTA, and washed as above. Viability of cells recovered in suspension was estimated from their ability to exclude 0.2% trypan blue. Differential counts were performed on cytocentrifuge preparations stained with Diff-Quik.
Assay of HzOz and & Secretion by Adherent CelLs-The oxidation of scopoletin by HzOz, catalyzed by horseradish peroxidase, was measured fluorometrically as described (23). Triplicate coverslips with adherent macrophages were rinsed by agitation in 4 beakers of 0.9% NaCl, excess fluid was drained off, and the slips were transferred to 16-mm wells containing 1.5 ml of KRPG with 17 p M scopoletin, 0.44 purpurogallin unit/ml of peroxidase, and 100 ng/ml of PMA. After 90 min of incubation at 37 'C in a water bath, the fluorescence of the supernatant was recorded in a Perkin-Elmer MPF44A fluorometer with excitation 350 nm and emission 460 nm. For 0; release, the 1-ml reaction mixture contained 100 p~ ferricytochrome c and 100 ng/ml of PMA in KRPG. The supernatants were cleared by centrifugation and the absorbance recorded at 550 nm. After subtracting the absorbance of supernatants from cell-free wells, cytochrome reduction was calculated using E = 21.0 m"' .
Matched triplicate coverslips were rinsed in the same way and lysed in 0.3 ml of 0.5 N NaOH for determination of the protein content by the method of Lowry et al. (24) using bovine serum albumin as a standard. ROI secretion by adherent cells is expressed as nanomoles of product/90 min/mg of cell protein.
Assay of Superoxide Dismutuse-Superoxide dismutase (EC 1.15.1.1) was measured in macrophages which were untreated or exposed to DDC for 60 min at 37 'C in MEM. The cells were rinsed 4 times in 0.9% NaC1, exposed to 0.2% (w/v) of Triton X-100 for 30 min at 4 "C, and assayed as described by McCord and Fridovich (25), expressing activity as Ananomoles of cytochrome c reducedlminlmg of protein.
Assay of NAD(P)H Oxiduse-Ferricytochrome c was acetylated by the procedure of Kakinuma and Minakami (21) with minor modifications. 300 mg of ferricytochrome c in 10 ml of half-saturated sodium acetate was slowly mixed with acetic anhydride (200 nmol/mol of cytochrome) and stirred for 40 min at 0 "C, followed by 24-h dialysis against distilled water. The product was stored at -80 "C in 0.9% NaCl. The efficiency of the acetylated compound in detecting 0; released from intact PMA-stimulated cells was 50% of that of native cytochrome c, which was taken into account in calculating results. Based on the modification by Bellavite et al. (15) of the assay of Curnutte et al. (26), 2.5 X lO6-3.O X lo6 cells were suspended in 1 ml of KRPG with 30 p M acetylated cytochrome c and 2 mM NaN3 ( t o inhibit cytochrome oxidase) in a 1-cm light path cuvette with an airdriven stirrer (27) in the thermostatted (37 "C) chamber of a Perkin-Elmer 557 dual wavelength spectrophotometer continuously recording AA (550-540 nm) ( E = 19.1 m"' ).
Unless indicated otherwise, the reaction was started by the addition of 100 ng/ml of PMA, stopped during maximum velocity by the addition of 0.05% (w/v) deoxycholate, and restored 45 s later by the addition of various concentrations of NAD(P)H. All rates reported with NADPH as substrate were completely abolished by 270 units/ml of superoxide dismutase. Activity was expressed as nanomoles of cytochrome c reduced/min/106 cells.
Oxidized-Reduced Difference Spectra-Spectra from 2.5 X 10'-8.0 X 10' cells in 1 ml of KRPG were stored in the computer of the Perkin-Elmer 557 spectrophotometer. After addition of a few grains of dithionite to the same cuvette, the spectra were redetermined and subtracted automatically from those stored.

Characteristics of the Cell Populations-
The adherent peritoneal cells from untreated mice or mice injected with NaI04 (28), sodium caseinate, or proteose peptone were recovered in suspension for enzymatic studies. As shown in Table I , morphologically identifiable macrophages comprised 88-94% of the cells and neutrophils 50.2%. In contrast, elicited peritoneal granulocyte populations contained 94% granulocytes and -5% macrophages. Periodate-or caseinate-elicited macrophages released an average of 7.4 times more H202/90 min/ mg of protein than resident or proteose-peptone-elicited cells. After overnight incubation, during which their Hz02 secretory capacity declined by about half, the activated macrophages released 16-22% as much H202 as freshly harvested granulocytes.
Intraperitoneal injection of inflammatory agents might elicit the immigration of different subsets of macrophages than those already present. Therefore, to compare activated and nonactivated macrophages derived from one population, we exposed resident peritoneal macrophages to NaI04 in vitro. As shown in Fig. 1, left, 3-day exposure to 0.5 mM NaI04, the highest concentration which did not reduce the amount of cell protein adherent to the coverslips, resulted in Hz02releasing capacity 23 times greater than that of control cells. Granulocytes were virtually undetectable in these in vitroactivated populations.
The marked difference between resident and activated cells was preserved over PMA concentrations ranging from 10-103 ng/ml (Fig. 1, right). Negligible ROI secretory activity was observed in cells exposed to vehicle without PMA, as illustrated in Fig. 1, right. The concentration of PMA giving a half-maximal response was 4 ng/ml. .Subsequent studies used the optimal concentration, 100 ng/ml.

Use of DDC to Enhance Detection of Superoxide Secreted by Intact Macrophages-If
HzOz arises from the dismutation of 0; and if both ROI can be detected outside the cell with equal efficiency, then twice as much 0; should be detected as HzOz (25). In fact, in the experiment illustrated in Fig. 2, 17 times more H202 was detected than 0; secreted by the same cells in response to PMA. We speculated that endogenous superoxide dismutase might be catalyzing the dismutation of 0; to H202 before 0; could react with ferricytochrome c. If so, then inhibition of Cu-,Zn-dependent superoxide dismutase with the copper chelator DDC (19) might favor the detection of 0; at the expense of H202. This was observed. Exposure of  activated macrophages to 1-20 mM DDC for 1 h inhibited endogenous superoxide dismutase activity by approximately 70% (Fig. 2, left). These concentrations were nontoxic, as judged by adherent cell protein (data not shown). In response to PMA, macrophages pretreated with 10 mM DDC released 15 times more 0; (Fig. 2, center) and 3 times less H202 (Fig.   2, right) than cells not exposed to the chelator. In fact, after DDC treatment, macrophages released almost exactly as much 0; as the H202 they released without exposure to DDC.
Superoxide Generation by Cell Lysates-For a precise comparison of 0; release by intact and lysed cells, we used a modification of the method of Bellavite et al. (15), in which 0; production is monitored continuously in the same cuvette before and after lysis of the cells. We first used native ferricytochrome c. Intact macrophages at rest reduced ferricytochrome c at a negligible rate (Fig. 3, left). After addition of PMA, 0.64 nmol of ferricytochrome c was reduced/min/106 macrophages. The addition of 0.05% (w/v) deoxycholate promptly abolished this response. The further addition of 1 2.8 0.6 3.2 1.4 0.0 2.9 10.8 mM NADPH resulted in renewed reduction of ferricytochrome c. However, the rate was greater than that with intact cells. Moreover, addition of superoxide dismutase had little effect (Fig. 3, left), suggesting that most of the reduction by lysates may have been due to cytochrome reductase rather than to 0;. In contrast, when acetylated ferricytochrome c was used (20,21), the rate of cytochrome reduction upon addition of 1 mM NADPH to the lysate was 100% of the value with intact cells and was decreased to 0 by the addition of 270 units/ml of superoxide dismutase (Fig. 3, right).
The ability of exogenous NADPH to restore 0; production declined rapidly as a function of time after exposure of the cells to deoxycholate (tH -90 s) (Fig. 4). Addition of bovine serum albumin, EDTA, glycerol (29), or N-a-p-tosyl-L-lysine chloromethyl ketone HC1 (29) did not prolong the interval after which NADPH could be added with full effect (data not shown). However, following the addition of NADPH, the rate of 0; production was constant for -2 min.
Effect of Detergents and pH-The ability to restore 0; production by lysed cells to 100% of the value for intact cells was critically dependent on both the choice of detergent and its concentration. In contrast to a report with human granulocytes (30), polyoxyethylene-12-tridecyl ether (Renex 30) was ineffective, as was polyoxyethylene-10-tridecyl ether, Triton X-100, and Nonidet P-40 (Table 11). With deoxycholate, 0.025% did not lyse cells, whereas 0.1% lysed cells but did not permit reconstitution of their capacity to release 0;. At 0.05%, deoxycholate both lysed cells completely and permitted full reconstitution (Fig. 5). With sonication of cells, 511% as much 0; was produced after addition of NADPH as was seen with deoxycholate lysis (data not shown).
Full reconstitution of 0; production after detergent lysis was also critically dependent on the pH of the buffer. As shown in Fig. 6, release of 0; from intact DDC-treated cells was almost nil at pH I 7.0 and rose steeply to a maximum at pH 2 8. 0; production by lysed cells was optimal at pH -7.4, and at this pH, the values seen with intact and lysed cells were virtually the same. If the cells were not pretreated with DDC, then the 0; detected from lysed cells exceeded that from intact cells at all pH <8 (Fig. 6), possibly reflecting the dilution of endogenous superoxide dismutase upon cell lysis.
Based on the above results, kinetic analyses were carried out at pH 7.35, using DDC-pretreated cells lysed with deoxy- phages activated in uiuo with NaIOa were suspended in 1 ml of KRPG containing 2 mM NaN3 in a dual wavelength spectrophotometer at 37 "C and exposed sequentially to 100 ng/ml of PMA, 0.05% (w/v) deoxycholate (DOC), 1 mM NADPH, and 270 units/ml of superoxide dismutase (SOD). cholate at 0.05%, followed 45 s later by the addition of various amounts of NADPH or NADH.
Kinetic Studies of NAD(P)H Oxidase-Reduction of acetylated ferricytochrome c by lysates of PMA-stimulated macrophages or granulocytes was 6-14 times faster following addition of NADPH than after the addition of equivalent concentrations of NADH (Table 111). The preference of the 0;-generating enzyme for NADPH was greater than reflected by these figures, because 100% of the cytochrome reduction seen after addition of NADPH, but not after addition of , or granulocytes (0) was recorded in duplicate samples after adding 1 mM NADPH at the indicated times following the addition of 0.05% deoxycholate. Peak activity was seen when NADPH was added 30-60 s after the detergent and was 100% of the rate before detergent. The data are expressed relative to the rate observed at 30 s. All cells were DDC-treated before assay. Tine (min) NADH, was abolished by superoxide dismutase (data not shown). 0; production by lysates in the presence of NADPH was strictly dependent on stimulation of the phagocytes with PMA prior to detergent lysis ( Table 111). Addition of PMA after lysis had no effect (data not shown).
The reciprocal of the rate of 0; production by lysates of PMA-stimulated cells was a linear function of the reciprocal of NADPH concentration (Fig. 7), permitting estimates of the K,,, and V,, of the oxidase (summarized in Table IV). Surprisingly, the V,, of resident macrophages and J774G8 histiocytoma cells, which secrete extremely little 02, was similar to that of macrophages activated with NaI04 or caseinate. The V,, for granulocytes was 2.3-3.0 times as great as that for the various macrophage populations. In contrast, the K,,,

Effect of detergents on superoxide production by macrophage lysates
The lysates were adherence-purified macrophages from NaI04treated mice triggered with 100 ng/ml of PMA. * Release of 0; by macrophages in the presence of the indicated concentration of detergent with and without 1 mM NADPH is given as a per cent of the value without detergent. The latter averaged 1.1 nmol/min/1O6 cells.

FIG. 5.
Effect of concentration of deoxycholate on 0 2 production by macrophage lysates. Adherence-purified, DDCtreated, NaI0.-activated macrophages were assayed as in Fig. 3, except that the concentration of deoxycholate was varied as shown (per cent w/v). 0, maximal rates of 0; production before addition of NADPH. Only 20.05% deoxycholate resulted in abolition of 0; production and lysis of all the cells as judged by microscopic examination. Columns indicate 0; production after addition of 1 mM NADPH. Values are expressed as per cent of the rate before addition of detergent (1.1 nmol/min/106 cells).
values of activated macrophages and granulocytes were similar and stood apart from the greater values seen with nonactivated macrophages. Thus, compared to periodate-activated macrophages, the K, for proteose-peptone-elicited macrophages was 4.3-fold higher, for resident macrophages 8.7-fold higher, and for J774G8 cells 30-fold higher.
Cytochrome b559 Content-In granulocytes, a unique b-type cytochrome is involved in the generation of 02, and its deficiency results in extremely low levels of release of ROI (6). Such a cytochrome has been detected in human mononuclear phagocytes (31). T o determine whether differences in content of this cytochrome might correlate with the ROI secretory capacity of variously activated macrophages, we recorded I"-without DDC pH pH FIG. 6. Effect of pH on 0; generation by intact and deoxycholate-lysed macrophages. NaI0,-activated adherence-purified macrophages were (left) pretreated or (right) not pretreated with DDC as described under "Materials and Methods." They were resuspended in 0.9% NaC1,5.5 mM glucose, and diluted into KRPG of the indicated pH for assay of 0; production before (0) and after (A) addition of deoxycholate and NADPH as in Fig. 3. Data are means of duplicates.
Intact granulocytes showed an a band at 559 nm, without a shoulder a t 550 nm (Fig. 8, left). Addition of 0.2% Triton X-100, or rescanning 4 min after adding dkhionite, led to the appearance of a large peak at 474 nm, presumably due to myeloperoxidase (Fig. 8, left). Based on the reported extinction coefficient for the cytochrome b559 of human blood neutrophils (32), mouse peritoneal granulocytes appeared to contain 115 -+ 6 pmol of cytochrome b559/mg of protein or 5.9 f 0.3 pmol/106 cells (mean -+ S.D., n = 8), virtually the same as reported for normal human blood neutrophils (10). In contrast, activated macrophages showed major bands at 550 nm and at 559, and no band at 474 (Fig. 8, center). In nonactivated macrophages, the band at 559 nm equaled or exceeded that at 550 nm (Fig. 8, right). The findings at 550 nm are consistent with the relative abundance of mitochondria in macrophages compared to granulocytes. The results at 474 nm suggest that contamination of the macrophages with myeloperoxidasecontaining cells (granulocytes or monocytes) was negligible. Of chief interest, differences in the peak at 559 nm among the macrophage populations bore no relation to the capacity of the cells to secrete ROI. For example, proteose-peptoneelicited macrophages, which released scant 02, appeared to contain 108 f 10 pmol of cytochrome bsS9/mg of protein or 7.6 -+ 0.7 pmol/106 cells ( n = 5) (as much as granulocytes, and far more than activated macrophages).

DISCUSSION
Two aspects of the respiratory burst of macrophages lend particular interest to a study of its enzymatic basis. First, this pathway can account for an important portion of the antitumor, antimicrobial, and inflammatory potential of the macrophage (2, 3). Second, the respiratory burst of the macrophage is highly sensitive to regulation. Thus, interferon-?, a T lymphocyte product, enhances ROJ secretory capacity in both human (18) and murine' macrophages, while a tumor cell product has the opposite effect (23). That the enzymatic basis of ROI secretion in macrophages has been much less studied than in granulocytes can be attributed in large part After exposure to PMA or dimethyl sulfoxide vehicle alone, cells were lysed with 0.05% deoxycholate. When 0; release had ceased, the indicated concentration of nucleotide was added and the renewed rate of 05 production recorded as nanomoles/min/1O6 cells. The rate of 05 release by intact cells in each experiment was the same as the highest value obtained after the addition of NADPH to the lysates. macrophages to secrete ROI may be due to an increased affinity of their oxidase for NADPH. We found no evidence that the specific activity of the oxidase was higher in activated than in resident macrophages.
It was first necessary to resolve the problem that 0; production by intact cells stimulated with PMA appeared to be much lower than previously reported (33) and than expected from simultaneous measurements of HzOz. This discrepancy has been noted before with both macrophages (23, 34) and The scan performed before adding dithionite was automatically subtracted from that obtained after adding dithionite, and the difference is displayed. The scale for absorbance units is bracketed between vertical arrows and for wavelength between horizontal arrows. k f t , peritoneal granulocytes (5 X io7 cells).
The upper trace was recorded immediately after adding dithionite and the lower trace 4 min later. Center, macrophages activated by injection of caseinate (5.5 X lo7 cells, upper trace) or NaI04 (8 X 1 0 ' cells, lower trace). Right, proteose-peptone-elicited macrophages (4.6 X IO7 cells, upper trace) and 5714G8 histiocytoma cells (2.5 X lo7 cells, lower trace). granulocytes (35). The present results suggest two explanations. First, the relative recoveries of 0; and H202 from the same cell populations varied inversely from each other in response to the concentration of DDC to which the cells had been exposed, over the same range in which endogenous superoxide dismutase was partially inhibited by the chelator (19) (the residual superoxide dismutase activity may have been mitochondrial). That is, the more superoxide dismutase was inhibited, the more 0: and the less HzOz was detected. This suggests that endogenous superoxide dismutase reacted with 0; more efficiently (rate constant, 2 X lo9 M" s-' (36)) than did exogenous ferricytochrome c (rate constant, 5 x 10' " I s-l (37)). However, we could detect no superoxide dismutase in the extracellular medium (data not shown). The simplest explanation is that 0: was produced intracellularly. In the absence of DDC, most of the 0; reacted with cytosolic superoxide dismutase to form HzOz, much of which diffused from the cell. Possible intracellular sites of formation of 0; include the inner surface of the plasma membrane or vesicles derived from it. A stoichiometric analysis with granulocytes (35) and a cytochemical study with macrophages (38) support the idea that HzOz may arise intracellularly. However, the loci of ROI generation in macrophages and the mechanism of the DDC effect are not defined by our experiments, and additional approaches will be required to test these hypotheses.
A second factor dramatically affecting apparent rates of secretion of 0; was the pH of the assay buffer. For example, with increasing alkalinity over a range as narrow as from pH 7.25 to 7.50, detectable 0; increased 4-fold. Indeed, increasing alkalinity would be expected to retard the reaction, 20; + 2H+ + Hz02 + 0 2 (25). However, it seems unlikely that pH would be markedly affected at the intracellular site where most dismutation seems to be occurring. The mechanism of the pH effect is unknown. In any case, the use of slightly more alkaline media in other studies, especially those in which volatile buffers were employed, together with possible variations in cellular superoxide dismutase content, may help explain reported differences in macrophage 0; secretory rates.
After optimizing conditions for the secretion of 0: by intact cells, we next endeavored to optimize the production of 0; by lysed cells. The comparison was aided by the method of Bellavite et al. ( E ) , in which the same cell population is monitored in a cuvette before and after lysis. We compared sonication and 5 detergents, each under a wide range of conditions such as buffer composition (data not shown) and detergent concentration. The only condition permitting completely lysed cells to produce 100% as much 0; as they did before lysis was the use of 0.05% deoxycholate. Even so, if the addition of NADPH was delayed by more than 90 s after cell lysis, the capacity to generate 0; had already declined by 50%. Lability of the oxidase has also been noted after subcellular fractionation of human granulocytes (29,39). In the case of mouse peritoneal leukocytes, the marked degree of this lability makes both subcellular localization and purification of the oxidase extremely difficult. However, the system was suitable for kinetic 3tudies on unfractionated lysates.
Finally, macrophage lysates contained a superoxide dismutse-resistant principle which rapidly reduced ferricytochrome c in the presence of NADPH, presumably NADPHdependent cytochrome reductase. Our studies would not have been possible without the use of acetylated cytochrome c (20,21) to minimize this reaction while preserving the susceptibility of cytochrome c to reduction by 0:. Reduction of acetylated cytochrome c by lysates of PMA-triggered macrophages in the presence of NADPH could not be attributed to radical production by cell-associated DDC, because the K,,, for NADPH was the same using cells not exposed to DDC (data not shown). Moreover, negligible 0; production was seen if NADPH was omitted or if PMA was added after rather than before lysis.
There are few previous reports dealing with the kinetics of the macrophage respiratory burst in cell-free preparations. The classic studies of Romeo et al. (14) and more recent work from the same laboratory (15,16) did not involve direct comparisons among macrophages in different states of activation. In separate studies with resident or casein-elicited macrophages from the peritonea and lungs of guinea pigs and rabbits, these workers found K,,, values for NADPH ranging from 0.03 to 0.72 mM (14-16). The influence of immunologic activation on the oxidative metabolism of guinea pig or rabbit macrophages has not been explored as thoroughly as in mouse or man. Only Sasada et al. (17), working with mouse peritoneal cells, directly compared the kinetics of the oxidase in cell-free preparations from resident and activated macrophages. The latter, taken from mice injected with bacterial lipopolysaccharide, displayed a 45% increase in VmaX and a 1.9-fold decrease in K,,, for NADPH (from 0.094 to 0.049 mM). However, in that study, the cells were disrupted by sonication. The activity recovered was not compared to the initial activity but can be calculated to have been ~4 . 7 % and probably <0.47% of that of the intact cells. Nonetheless, after overnight culture, the results using that technique (17) were similar to those reported here.
As yet we have no explanation for the markedly enhanced affinity of the oxidase of activated macrophages for NADPH compared to that of resident macrophages. Differences in the degree of triggering of the oxidase by PMA seem unlikely, since PMA receptor number and affinity do not differ between activated and resident macrophages (40), and increasing the PMA concentration did not increase the response of nonactivated cells. An inhibitor of the oxidase (4, 41) may be diminished in activated macrophages. A new oxidase with a lower K,,, may be induced by activation, or the existing oxidase may undergo allosteric changes. Activated macrophages may assemble a multicomponent oxidase more efficiently, such as by incorporating more cytochrome b559 or ubiquinone, which might lower the apparent affinity of the electron transport chain as a whole for NADPH. We did not observe a correlation between cytochrome b559 content and capacity to secrete ROI.
However, the contribution of mitochondrial cytochromes to the spectra may have obscured critical variations. Such differences should be sought in membrane-rich fractions free of mitochondria.
The change we observed in the K,,, of the oxidase for NADPH as resident macrophages (0.41 mM) became activated (0.05 mM) seems likely to be physiologically relevant, when compared with the reported intracellular NADPH concentration in resident and activated macrophages (0.14 and 0.30 mM, respectively, before PMA stimulation, and 50% lower after PMA stimulation (17)). Thus the data provide support for the hypothesis that differences in ROI secretory rates among macrophage populations primarily reflect differences in production of ROI rather than in the proportion of ROI released into the medium, accumulated, or catabolized. It will now be of interest to compare the changes induced by activation to those induced by deactivation, such as occurs upon exposure of macrophages to factors derived from tumors and some nonmalignant cells (23).